> If you consider any of the vba-m code, this becomes less of a mystery.
Ah. Reminds me of Snes9X. The code itself isn't horrible, but there's absolutely no structure to it. CPU serialization, PPU rendering, ROM loading, an embedded ROM, BIOS HLE, DMA transfers, system reset, really long functions, a function called CPULoop that does colorspace conversion blitting, etc ... weighing in at over 4,400 lines, all in a single source file. I would add virtually no comments, but I'm not really great at that myself either.
Sadly, I'd say that most emulators are written this way. If you ever want to really wreck your brain, take a look at the ZSNES source. It's all of the above, in x86 assembly (no code flow indentation), with zero comments, and cryptic names (often just registers.)
I have my warts as well, of course. Shared processor cores means the ARM logic isn't in the gba/ project, so that it can be shared by the ST018 SNES CPU. And I tried to do this operator overloaded thing for the MMIO registers that just totally didn't work out at all like I wanted. Undoing that is high on my todo list. If you do have any constructive criticism on my code, I'd be happy to improve upon it, time permitting.
> The fortunate thing here is that any computer capable of running dolphin can do (at least one instance of) higan, no problem.
Oh, I mean the simulation is too slow. So parts of some games will feel like they are slowing down, even though you are at 60fps output.
> See, it's a good thing that dolphin doesn't use all your cores!
> The ultimate test will be 4 Higans, Dolphin on LLE audio, and recording it at the same time!
From the way I see it, a PC fast enough to run Dolphin should be no problem for higan.
higan is entirely single-threaded, so you'll basically need a system that has one core for each instance of higan, plus however many more you need for Dolphin. It would probably make sense to use the scheduling APIs to make sure higan and Dolphin don't try and share a core at the same time.
So ... 4x higan + Dolphin will probably want something like a Core i7-3930K or better.
1x higan + Dolphin should be okay on a more modest i7-2600K, presuming that the two don't end up fighting each other for a shared resource.
Ah. Reminds me of Snes9X. The code itself isn't horrible, but there's absolutely no structure to it. CPU serialization, PPU rendering, ROM loading, an embedded ROM, BIOS HLE, DMA transfers, system reset, really long functions, a function called CPULoop that does colorspace conversion blitting, etc ... weighing in at over 4,400 lines, all in a single source file. I would add virtually no comments, but I'm not really great at that myself either.
Sadly, I'd say that most emulators are written this way. If you ever want to really wreck your brain, take a look at the ZSNES source. It's all of the above, in x86 assembly (no code flow indentation), with zero comments, and cryptic names (often just registers.)
I have my warts as well, of course. Shared processor cores means the ARM logic isn't in the gba/ project, so that it can be shared by the ST018 SNES CPU. And I tried to do this operator overloaded thing for the MMIO registers that just totally didn't work out at all like I wanted. Undoing that is high on my todo list. If you do have any constructive criticism on my code, I'd be happy to improve upon it, time permitting.
> The fortunate thing here is that any computer capable of running dolphin can do (at least one instance of) higan, no problem.
Oh, I mean the simulation is too slow. So parts of some games will feel like they are slowing down, even though you are at 60fps output.
> See, it's a good thing that dolphin doesn't use all your cores!
> The ultimate test will be 4 Higans, Dolphin on LLE audio, and recording it at the same time!
From the way I see it, a PC fast enough to run Dolphin should be no problem for higan.
higan is entirely single-threaded, so you'll basically need a system that has one core for each instance of higan, plus however many more you need for Dolphin. It would probably make sense to use the scheduling APIs to make sure higan and Dolphin don't try and share a core at the same time.
So ... 4x higan + Dolphin will probably want something like a Core i7-3930K or better.
1x higan + Dolphin should be okay on a more modest i7-2600K, presuming that the two don't end up fighting each other for a shared resource.